Inferring the Driver's Lane Change Intention through LiDAR-Based Environment Analysis Using Convolutional Neural...

Area of Science:

• Artificial Intelligence
• Computer Vision
• Automotive Engineering

Background:

• Predicting driver behavior is crucial for developing advanced driver assistance systems (ADAS) and autonomous driving technologies.
• Understanding the surrounding environment and vehicle dynamics is essential for tactical driving maneuvers like lane changes.

Purpose of the Study:

• To develop and evaluate a Convolutional Neural Network (CNN) model for predicting driver lane change intentions.
• To adapt a CNN architecture for integrating both spatial and non-spatial driving data.
• To explore the application of lane change prediction in ADAS and autonomous vehicle decision-making.

Main Methods:

• A modified CNN architecture was designed to process spatial (surrounding environment) and non-spatial (relative speed, etc.) input variables.
• The CNN model was trained and tested within a driving simulation environment.
• Performance was evaluated by comparing the model's accuracy against random guessing and its ability to differentiate driving profiles.

Main Results:

• The proposed CNN model demonstrated a higher accuracy in predicting lane changes than random chance.
• The model successfully captured subtle behavioral differences among various driving profiles.
• The integrated approach effectively utilized both environmental and vehicle kinematic data.

Conclusions:

• CNNs are effective tools for modeling and predicting driver lane change behavior.
• Anticipating lane changes can significantly enhance the safety and human-like performance of ADAS and autonomous vehicles.
• The developed model provides a valuable data source for improving vehicle decision-making algorithms.